Outline of this work. The following three chapters introduce different topics that underlie the linguistic data discussed in the bulk of this dissertation. Chapter 2 presents an overview of Shiwiar grammar from a typological perspective and it explains the relationship of Shiwiar to the rest of the Chicham languages, as well as to other unrelated languages spoken in the area. Chapter 3 introduces the Shiwiar people, their history and their culture. It also includes observations about the use of language in Shiwiar communities from a sociolinguistic standpoint. Chapter 4 is a detailed description of the data sources, the methodology used for data collection, and information about the primary data that this thesis is based on. The phonology of Xxxxxxx is discussed in Chapter 5, including an overview of the phonological inventory, insights on syllabic structure and phonotactics and a description of elision processes that have had substantial effects on Shiwiar morphophonology. This chapter also examines the phonetic characteristics of speech sounds in Shiwiar. An overview of Shiwiar morphology, syntax and information structure is presented in Chapter 6. In this chapter, all Shiwiar word classes are listed and defined. Then, the highly synthetic nature of Shiwiar morphology is introduced. Syntactic topics, such as constituent order, finiteness and clause combining, are also examined. Finally, the marking of information structure is discussed. Nominals are the focus of Chapter 7. First, the characteristics of nouns, adjectives and pronouns are described, as well as the differences between them. Then, derivational and inflectional morphology is presented, including possession and case marking. This chapter also addresses the internal syntax of noun phrases and the use of demonstratives and other nominal modifiers.
Outline of this work. In this dissertation we develop new, open source Java software for scientific com- puting and image processing with an emphasis on solving large-scale inverse problems. The rest of this work is organized as follows. Chapter 2 describes math- ematical theory and computational methods for solving two important classes of image processing algorithms: deconvolution and super-resolution. In Chapter 3 we motivate the choice of Java for our imaging applications and describe a high- performance Java library for scientific computing and image processing, which we call Parallel Colt. Chapter 4 discusses the details of the implementation, usage examples and benchmarking of various image processing algorithms implemented on top of Parallel Colt. Conclusions are summarized in Chapter 5. In Appendix A we describe the derivation of a fast Fourier transform algorithm and Appendix B provides information about software packages that use the libraries and plugins described in this work.
